AN INTELLIGENT NETWORK INTRUSION DETCETION SYSTEM USING DATA MINING AND KNOWLEDGE BASED SYSTEM

摘要

network attacks that incorporates a knowledge based system and data mining techniques. To extract hidden knowledge from KDDCup’99 dataset, hybrid data mining process is used. The intrusion dataset for the study is collected from MIT Lincon lab. A predictive model is constructed on total datasets of 63, 661 instances using JRip rule induction, Na?ve Bayes,J48 decision tree and Multilayer Perceptron (MLP) Neural Network. During training 99.91% prediction accuracy is achieved by J48 decision tree. So, the J48 model is integrated with knowledge based system automatically for designing intelligent network intrusion detection and prevention system. In addition, knowledge is acquired, represented and organized in the knowledge based so as to suggest possible prevention for detected attacks. Evaluation results show that the proposed system registers 91.43% accuracy in network intrusion detection and 85% in user acceptance testing. This indicates that the performance of the proposed system is promising to design an intelligent network intrusion detection system that can effectively predict and provide a prevention mechanism. The system cannot update the knowledge of prevention techniques automatically which need further researches.